ID Number		160
Study Type		Human / Provocation
Model		450, 900 MHz (NMT/CW, GSM, Tetra) exposure to adults and children and analysis of cognitive function and memory
Details		Human subjects (n=36 adults, 2 groups of 18) were exposed to 915 MHz (FM, GSM) signals from a modified mobile phone emitting 1 watt of power and mounted on a headset exposing the left temporal/tempo-parietal area for 25-30 minutes during which a series of cognitive function and memory tests (Cognitive Drug Research [CDR] battery) were administered. The initial study (1999) reported a decrease in choice reaction time of 15 msec (~4%) vs controls with FM but not GSM RF exposure. Neither FM nor GSM exposure had any effect on related attention tasks or on any memory function. The authors interpreted the results as a possible effect due to mild RF heating on the angular gyrus in the left temporal region of the brain. An extension of this study was presented at the 21st Bioelectromagnetics Society Meeting in Long Beach, CA and reported, in addition to the effect on choice recognition time, an increase in cerebral blood flow using a Logidop 3, which is a trans-temporal 2 MHz ultrasound probe positioned coaxially with the middle cerebral artery (as described in Fujioka, 1998, J Vascular Technol, vol (?):121-125). In a related study, volunteers were exposed to 450 MHz TETRA signals (4 watt peak power output - no SAR listed) and tested as above. In this study, no effects were reported using the CDR test battery. Further studies performed in children (n = 18, age 10-12 yrs) during a 30-35 minute exposure to 902-MHz GSM RF at various power levels (peak SAR of 0.28 W/kg as determined using a child phantom model) reported no statistically significant effects on any of the 22 measures of reaction time, accuracy, and a sensitivity index (corroborating reports by Haarala et al, 2005), although the authors did acknowledge a trend toward reduced reaction time in the children that could be related to their earlier reported findings in adults. They also emphasize that the findings in adults were with more powerful analogue signals, and the exposure to children was only with digital (GSM) signals with a lower average SAR. The authors conclude that there is no evidence to suspect children might be more sensitive to RF exposure from mobile phones than adults given the findings of the current study.
Findings		Effects
Status		Completed With Publication
Principal Investigator		University of Bristol, UK - a.w.preece@bristol.ac.uk
Funding Agency		DOH, UK
Country		UNITED KINGDOM
References		Preece, AW et al. Bioelectromagnetics., (2005) 26 Suppl 7:S138-S143 Preece, AW et al. Int. J. Radiat. Biol., (1999) 75:447-456
Comments		The CDR battery of tests included analysis of 1) simple reaction time, 2) choice reaction time, 3) vigilance reaction time, 4) vigilance accuracy, 5) choice reaction time accuracy, 6) spatial memory sensitivity index, 7) memory scanning sensitivity index, 8) delayed recall sensitivity index, 9) immediate recall accuracy, 10) delayed recall accuracy, 11) delayed picture sensitivity index, 12) spatial memory reaction time, 13) memory scanning reaction time, 14) delayed recall reaction time, and 15) delayed picture reaction time (the first 5 tests analyzing reaction time and accuracy of attentional tasks, and the rest being measures of memory function). Previous studies have shown normal inter-subject variation in choice reaction time is on the order of ~40 msec, within the range of exposed vs sham data in the 1999 Preece et al study. In comparing the initial studies by Preece from adults (Int. J. Rad. Biol, 1999, 75:447-456) with Kovisto et al (NeuroReport (2000) 11:1641-1643) and Jech et al (Bioelectromagnetics 2001): In the study by Koivisto, there are some positive findings specific for 3/14 tasks that measure attention and mental arithmetic, but when compared to study by Preece in adults, the positive test endpoints in one do not correlate with the same endpoint in the other. Preece used Cognitive Drug Research (CDR) software and did a 3-way comparison between digital, analogue, and sham exposed individuals with the following results: 1) simple rxn time - no effect 2) choice rxn time - effect with analogue only, p = 0.003 3) vigilance rxn time - no effect 4) vigilance accuracy - no effect 5) choice rxn time accuracy - no effect 6) spatial memory sensitivity index - no effect 7) memory scanning sensitivity index - no effect 8) delayed sensitivity index - no effect 9) immediate recall accuracy - no effect 10) delayed recall accuracy - no effect 11) delayed picture sensitivity index - no effect 12) spatial memory rxn time - no effect 13) memory scanning rxn time - no effect 14) delayed recall rxn time - no effect 15) delayed picture rxn time - no effect Koivisto used a test software by CogniSpeed (similar, but not the same) and compared only 902 MHz (GSM) digital and sham exposed individuals, and obtained the following results: 1) simple rxn time - small decrease in rxn time with RF exposure, p = 0.026 2) two choice rxn time - no effect 3) ten choice rxn time - no effect 4) subtraction task - no effect 5) subtraction time - decrease in time needed in subtraction task, p = 0.044 6) sentence verification - no effect 7) vigilance time - no effect 8) vigilance score - decrease in false alarms with RF exposure, p = 0.001 9) shape detection - no effect 10) object detection - no effect 11) object familiarity detection - no effect 12) semantic picture categorization - no effect 13) semantic word categorization - no effect 14) object name retrieval - no effect While the different battery of tests makes direct comparison difficult, the Koivisto study used tests more specific to cognitive function while Preece was looking at memory function as well. Preece found a decrease in choice rxn time with analogue RF exposure only, Koivisto did not use analogue signals, so no comparison can be made here. The effects observed by Koivisto on simple rxn time and vigilance, however, were not corroborated by the data from the Preece study using similar functional tests. The observation on subtraction time by Koivisto was not directly addressed in the Preece study. In addition to the studies not supporting each other, there is no concomitant support from the tests within the same study. For example, if vigilance (decrease in false alarms) decreased in the Koivisto study, one might expect vigilance time to also be affected, and it was not. Similarly if choice rxn time was affected in the Preece et al study, one might expect other parameters such as simple rxn time to also be affected, and they were not. Finally, neither study shows a robust effect on any parameter. While the p value for a decrease in choice rxn time in the Preece study is 0.003, the mean scores are 373.4 (analogue RF exposure – 0.07% difference), 384.4 (digital RF exposure), and 387.9 (sham RF exposure) and no raw data or standard deviation was given. In the Koivisto study, the score for simple rxn time was 282 +/- 39 (digital RF exposure) and 273 +/- 29 (sham RF exposure), for subtraction time was 245 +/- 148 (digital RF exposure) and 216 +/- 124 (sham RF exposure), and for vigilance was 517 +/- 48 (digital RF exposure) and 492 +/- 42 (sham RF exposure). The statistical methods were not described at all, but the overlap of SD's makes it hard to believe the p values were less than 0.05. In Jech et al, narcoleptics were measured specifically. In contrast to other studies such as Borbely et al (Neuroreport (2000) 11:3321-3325; Neuroscience Letters 275 (1999): 207-210) there was no effect on sleeping EEG. The authors did observe an increase in reaction time (20 milliseconds) when target stimuli were presented in the right side of a computer screen and only when subjects were sham exposed on the first day and RF exposed on the second day (not visa versa). This result is in contrast to the results of Eulitz et al (Neuroreport (1998) 9(14):3229-3232) who reported EEG changes but no ERP effects. The authors make comparison and claim their study supports Preece et al and Koivisto et al.